Hindawi Publishing CorporationExperimental Diabetes ResearchVolume 2012, Article ID 438238, 19 pagesdoi:10.1155/2012/438238

Research Article

TXNIP Links Innate Host Defense Mechanisms toOxidative Stress and Inflammation inRetinal Muller Glia under Chronic Hyperglycemia:Implications for Diabetic Retinopathy

Takhellambam S. Devi,1 Icksoo Lee,2 Maik Hüttemann,2

Ashok Kumar,1, 3 Kwaku D. Nantwi,1 and Lalit P. Singh1, 3

1 Department of Anatomy and Cell Biology, Wayne State University, Detroit, MI 48201, USA2 Center for Molecular Medicine and Genetics, Wayne State University, Detroit, MI 48201, USA3 Department of Ophthalmology, Wayne State University, Detroit, MI 48201, USA

Correspondence should be addressed to Lalit P. Singh, plsingh@med.wayne.edu

Received 14 July 2011; Accepted 27 November 2011

Academic Editor: Muthuswamy Balasubramanyam

Copyright © 2012 Takhellambam S. Devi et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Thioredoxin Interacting Protein (TXNIP) mediates retinal inflammation, gliosis, and apoptosis in experimental diabetes. Here,we investigate the temporal response of Muller glia to high glucose (HG) and TXNIP expression using a rat Muller cell line(rMC1) in culture. We examined if HG-induced TXNIP expression evokes host defense mechanisms in rMC1 in response tometabolic abnormalities. HG causes sustained up-regulation of TXNIP (2 h to 5 days), ROS generation, ATP depletion, ERstress, and inflammation. Various cellular defense mechanisms are activated by HG: (i) NLRP3 inflammasome, (ii) ER stressresponse (sXBP1), (iii) hypoxic-like HIF-1α induction, (iv) autophagy/mitophagy, and (v) apoptosis. We also found in vivo thatstreptozocin-induced diabetic rats have higher retinal TXNIP and innate immune response gene expression than normal rats.Knock down of TXNIP by intravitreal siRNA reduces inflammation (IL-1β) and gliosis (GFAP) in the diabetic retina. TXNIPablation in vitro prevents ROS generation, restores ATP level and autophagic LC3B induction in rMC1. Thus, our results show thatHG sustains TXNIP up-regulation in Muller glia and evokes a program of cellular defense/survival mechanisms that ultimatelylead to oxidative stress, ER stress/inflammation, autophagy and apoptosis. TXNIP is a potential target to ameliorate blinding ocularcomplications of diabetic retinopathy.

1. Introduction

Diabetic retinopathy (DR) is the most common cause ofblindness among the working age group people in the USand around the world. DR has long been considered as amicrovascular disease associated with vessel basement mem-brane thickening, blood retinal barrier breakdown, capillarycell death, acellular capillary, neovascularization, and retinaldetachment [1]. However, recent studies have demonstratedthat DR is a neurovascular disease that affects both theblood vessel and neuroglia [2, 3]. Chronic hyperglycemia-associated oxidative stress and low-grade inflammation are

considered to play critical roles in disease initiation and pro-gression of diabetic complications including DR [4–6]. Yet,the molecular mechanisms underlying hyperglycemic injuryand DR pathogenesis are poorly understood. Various glucosemetabolic defects and abnormal biochemical pathways areactivated in diabetes and under chronic hyperglycemia[7]. Recently, we demonstrated that thioredoxin interact-ing/inhibiting protein (TXNIP) is significantly increasedboth in the diabetic rat retina in vivo and in vitro in retinalendothelial cells in culture and causes pro-inflammatorygene expression for Cox-2, VEGF-A, ICAM1, RAGE, andsclerotic fibronectin (FN) [8–10]. Knockdown of TXNIP

2 Experimental Diabetes Research

by siRNA in the retina prevents early abnormalities of DRin streptozocin-induced diabetic rats, which include inflam-mation, fibrosis, gliosis and apoptosis [9, 10].

TXNIP is an early response gene highly induced bydiabetes and hyperglycemia [8, 11, 12]. TXNIP was ini-tially identified as one of the proteins that interacts withthioredoxin (TRX) and blocks its thiol reducing function[13, 14]. TRX is a 12-kDa protein, which scavenges ROSand maintains protein cysteine sulfhydryl groups by itsredox-active disulfite/dithiol sites. Recent findings furtherdemonstrate a potential role for TXNIP in innate immunityvia the NOD-like receptor-NLRP3/caspase-1 inflammasomeactivation and release of IL-1β in diabetes and oxidativestress [15, 16]. TXNIP interacts with NLRP3 leading to amultiprotein NLRP3 complex assembly and autoactivationof caspase-1. Activated caspase-1 in turn processes pro-IL-1β to its mature and active form leading to otherpro-inflammatory gene induction and inflammation [17,18]. Innate immune receptors TLR4 and RAGE and theirendogenous ligands such as high mobility group bindingprotein HMGB1 and S100 calgranulins are reported to beupregulated in diabetes [19, 20]. Currently, NF-κB is the onlypro-inflammatory transcription factor that is implicated inretinal inflammation in diabetes [3, 4, 8, 9] while the roleof other factors is undetermined. Innate immune receptorssignal through activation of NF-κB and AP1 transcriptionfactors to induce the expression of various pro-inflammatorycytokines and chemokines [4–6].

We recently showed that TXNIP mediates IL-1β expres-sion in primary Schwann cell in vitro as well as in vivo inpartial sciatic nerve injury [10] by the p38 MAPK/NF-κBpathway and transcription factor CREB. Similarly, TXNIPwas also shown to be responsible for pro-IL-β expression inadipocytes by high glucose [21] and HG-mediated NLRP3inflammasome and casapse-1 activation leads to IL-1β mat-uration [15, 16, 21]. In addition, chronic hyperglycemia mayalso evoke other innate host defense/survival mechanismssuch as ER stress/unfolded protein response (UPR) andhypoxic responses for cell viability in an inflamed andstressful environment [22, 23]. In this regard, chronichyperglycemia-associated TXNIP expression could poten-tially lead to thiol oxidation and misfolded protein accumu-lation in ER lumen and cause ER stress in diabetes. UPR is ahighly regulated intracellular signaling pathway that preventsaggregation of misfolded proteins. ER stress is known to bemediated by three UPR branches, namely, the IRE1-XBP1,PERK-eIF-1α, and ATF6 pathways [22, 23]. These UPRsignals provide an adaptive mechanism for proper proteinfolding and processing in the ER and establishment oftissue homeostasis and cell survival. However, when proteinmisfolding is not resolved, the ER stress and UPR triggeran apoptotic signal to remove demised cells by phagocytes.Therefore, ER stress may play a critical role in decipheringcell survival and premature demise in chronic disease andinjury including DR.

Currently, it is recognized that molecular abnormalitiesof DR begin early before clinically detectable pathologiesappear in the retina [2–4, 24]. The pathologies once set inmotion are not reversed even after glucose normalization

[25, 26]. Therefore, the detection of molecular abnormalitiesand cellular processes that ultimately lead to later pathologiesof DR is of utmost importance in developing therapeu-tic strategies to prevent blinding ocular complications ofdiabetes. Retinal Muller glia, blood vessels, and neuronal(ganglion) cells interact closely to maintain retinal tissuehomeostasis and cell survival [27]. Muller cells are theprimary glial type in the retina and along with microgliaand astrocytes are considered as resident innate immunecells. Under stress, they become activated and produce pro-inflammatory cytokines and growth factors to restore tissuehomeostasis [6]. However, in chronic diseases such as DR,retinal gliosis is prolonged, causing sustained inflammation,cell injury/death, and worsening disease. Nonetheless, a rolefor TXNIP in Muller glia reactivity, innate immune response,ER stress, and inflammation under chronic hyperglycemiaand diabetes has not been investigated before. In thisstudy, first, we demonstrate that TXNIP is significantlyinduced in the diabetic rat retina in vivo and mediates pro-inflammatory IL-1β expression and the induction of radialglial fibrillary acidic protein (GFAP) indicating Muller cellactivation. Secondly, we show that chronic hyperglycemiasustains TXNIP up-regulation in Muller glia in in vitroculture and orchestrates a temporal program of innate hostdefense mechanisms that lead to cellular oxidative stress,ER stress, inflammation, and autophagy/apoptosis. Thesefindings provide important insights as to how retinal Mullerglia might respond to chronic hyperglycemia in diabetes andthat TXNIP may play a critical role in glial dysfunction anddisease progression in DR.

2. Materials and Methods

2.1. Materials. Tissue culture media, serum and antibioticswere purchased from Invitrogen (Carlsbad, CA). Antibod-ies for TXNIP and beclin 1 were obtained from MBL(Woburn, MA). Anti-S-Nitroso-Cysteine (SNO-Cys) anti-body was from Sigma-Aldrich (St. Louis, MO). The LC3Bantibody kit for autophagy (Cat number L10382), Mito-Tracker Red CMXROS (M7512), ROS detection reagent CM-H2DCFDA (C6827), and ATP assay kits were from MolecularProbes (Invitrogen). Antibodies for pro-IL-1β, pro-caspase-1, NLRP3 (cryopyrin), and HMGB1 are from Santa CruzBiotechnology (Santa Cruz, CA), and VDUP1 (TXNIP),tubulin, and Actin were from Abcam (Cambridge, MA).Anti-p65 NF-κB and phospho-p65 NF-κB antibodies werefrom Millipore (Invitrogen). Electrophoresis Mobility ShiftAssay (EMSA) kit for Transcription factors XBP1 and HIF-1αto consensus DNA was custom-made by Signosis Inc. (Sun-nyvale, CA). Fluorescent-labeled secondary antibodies anti-rabbit and anti-mouse were obtained from Molecular Probeswhile those of the anti-goat antibodies were purchased fromAbcam (Cambridge, MA). Active caspase-1 (Green FLICACaspase-1) and Caspase-3 (Green FLICA caspase-3 & 7) werepurchased from Immunochemistry Technologies (Bloom-ington, MN). Predesigned TXNIP siRNAs were purchasedfrom Qiagen (SABiociences). TRIZOL for RNA isolationwas from Invitrogen and PCR primers were also synthesizedby Invitrogen or Applied Biosystems. First strand cDNA

Experimental Diabetes Research 3

synthesis kit and SYBR green reagents were purchased fromBiorad as well as from Applied Biosystems. Primer sequencesare available upon request.

2.2. Diabetes Induction of Rats. Diabetes of adult maleSprague-Dawley rats (∼275 g) was induced by intraperi-toneal injection of a single dose of streptozotocin, (STZ,65 mg/kg body weight, Sigma) dissolved in 0.01 M citratebuffer, pH 4.5 as described recently [8, 9]. The rats weretreated in accordance with the principles of NIH guidelinesfor the Care and Use of Laboratory Animals and approvedby the Institutional Animal Care and Use Committee.Intravitreal injection of siRNA targeted to rat TXNIPpromoter (transcriptional gene silencing, RNAi TGS) hasbeen described before [9] and performed in anaesthetizedrats with 40 mg/kg body weight (pentobarbital) on the righteyes of diabetic rats (Treatment) and a similar volume ofscramble-RNA on the left eye (Control). The injections wereperformed twice at day 23 and 27 and they were sacrificedat day 30. An overdose of pentobarbital (200 mg/kg weight)was given to euthanize the rats. The retina were removed,processed for immunohistological analysis or frozen imme-diately in liquid N2, and stored at –80◦C until used.

2.3. Cell Culture. We used a well-established rat retinalMuller cell line (rMC1) [28]. rMC1 cells were cultured inmedium containing DMEM (low glucose, 5.5 mM) andHam’s F-12 (3 : 1 ratio) supplemented with 5.0% fetalcalf serum (FCS) and 0.5 mg/mL gentamicin at 37◦C ina humidified chamber with a 5% CO2–95% air mixture[9, 29]. Cells at 70–80% confluence were replaced withlow serum overnight (0.2% serum medium). For the timecourse, HG (high glucose, 25 mM) with 24 h treatmentwas added first; then on the second day HG was addedat 4, 2 h, and 0 (no addition) to respective cultures beforeharvesting. Thus, all cells maintain a similar condition andlength of serum-deprivation during the entire experimentalprocedure. Similarly, for long-term cultures also, HG wasadded first to the 5-day culture, then after two days for 3days, 1 day, and 0 before harvesting. Media were changedevery 48 h. Cells were harvested by scrapping, snap frozenand stored at −80◦C until used.

2.4. SDS-PAGE and Western Blotting. Proteins were extractedin RIPA buffer containing protease inhibitors and theirconcentrations were determined using a Coomassie Plus(Bradford) Assay Reagent from Pierce (Product number23238) with BSA as the standard. Absorbance of blue dyeof the Coomassie-protein complex was measured at 595 nmusing a Gemini Microplate reader (Molecular Devices,Sunnyvale, CA). Thirty micrograms of protein extracts wassubjected to sodium dodecyl sulfate polyacrylamide gelelectrophoresis (SDA-PAGE) and Western blot analysis ofproteins was performed as described previously [30]. SantaCruz antibodies were used at 1 : 200 dilutions while anti-bodies from other sources were diluted at 1 : 1000 dilutions.Secondary antibodies were used at 1 : 3000 dilutions. ECLused to detect the immunoreactive bands.

2.5. Real-Time Quantitative PCR. Total RNAs were isolatedby TRIZOL method and first-strand cDNAs (from 1 μg RNA)were synthesized in 20 μL volume using the Bio-Rad iScriptcDNA synthesis kit [8, 12]. Messenger RNA expression wasanalyzed by real-time quantitative PCR using the Bio-RadChromo 4 detection system and SYBR Green PCR MasterMix from Bio-Rad or Applied Biosystems (Foster City, CA).Primers were designed using Primer Express v 2.0 (AppliedBiosystems) and synthesized by Invitrogen (Carlsbad, CA).Primers for the real-time Q-PCR will be available uponrequest.

The real-time PCR reaction mixture contained 1XSYBR Green PCR Master Mix, 400 nM forward and reverseprimers, and 2 μL cDNA in a final volume of 25 μL. The PCRcycling was programmed as 95◦C for 15 s, 55◦C for 30 s and72◦C for 30 s 40 cycles followed by the construction of amelting curve through increasing the temperature from 60◦Cto 95◦C at a ramp rate of 2% for 20 min. The real-time PCRsamples were evaluated using a single predominant peak as aquality control. Ct values were used to calculate the relativeexpression level of mRNAs that were normalized to actin[8, 12].

2.6. Determination of Cell Viability. We used MTT assayto measure cell viability at different time periods after HGaddition. MTT assay for cell viability was performed in48 well plates and with 0.5 mg/mL MTT in each well aspreviously described in our laboratory [31, 32]. rMC1 cells(1 × 104 cells) were grown to 70% confluence and serum-starved overnight and treated with HG. MTT was added for3 h at the end of the experiment, the media was removed,and cells were kept in 100 μL of DMSO for 10 minutes.The resulting color was diluted with 500 μL of distilled H2Oand detected at 570 nm using a Gemini Microplate reader(Molecular Devices, Sunnyvale, CA).

2.7. Intracellular Reactive Oxygen Species (ROS) Measure-ment. The formation of intracellular ROS in rMC1 cells wasdetected as described before [8, 12] by using the fluorescentprobe, 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluo-rescein diacetate, acetyl ester (CM-H2DCFDA). This dyecan enter living cells by passive diffusion it is nonfluorescentuntil the acetate group is cleaved off by intracellularesterase, and oxidation occurs within the cell. Approximately1 × 105 cells/mL were cultured in 24 well plates, serum-starved overnight, and glucose was added for the specifiedtime period. Then, CM-H2DCFDA (10 μM) was incubatedfor 60 min at 37◦C. The medium with the dye is aspirated(to remove the extracellular dye), washed with PBS (3x),and then the PBS is added to cells. The fluorescence wasmeasured in a Gemini Fluorescent Microplate Reader(Molecular Devices) with the bottom read scanning mode at480 nm excitation and emission at 530 nm.

2.8. Immunohistochemistry (IHC)

(i) Retinal Tissue. Immunohistology of retinal sections wassimilar to those described recently [9]. The retinas were fixed

4 Experimental Diabetes Research

in the eyecups with 4% paraformaldehyde in 0.1 M phos-phate buffer (PB) for 20 min. The retinas were cryoprotectedin a sucrose gradient (10%, 20%, and 30% w/v in PB, resp.).Cryostat sections were cut at 16 μm in Tissue Tek (OCTmounting medium). Retinal sections were blocked for 1 h ina PB solution that contained 5% Chemiblocker (Chemicon,Temecula, CA), 0.5% Triton X-100, and 0.05% sodium azide.The primary antibodies were diluted in the same solution(1 : 100 to 1 : 200 dilution depending on the antibody) andapplied overnight at 4◦C followed by appropriate secondaryantibodies (1 : 600 dilutions) conjugated with Alexa Fluor488 or Alexa Fluor 594 for 2 hours at room temperature.After washing with PB again, rMC1 cells were mounted withan aqueous mounting medium with anti-fade agent withDAPI and sealed with nail polish. The images were capturedby an OLYMPUS BX 51 fluorescence microscope, which isfitted with a triple DAPI/FITC/TRITC cube, a DP70 digitalcamera, and image acquisition software. Some images werealso captured with a Zeiss Apotome microscope with Z-section (Zeiss, Oberkochen, Germany). Similar magnifica-tion (400x) and exposure time were maintained throughoutfor comparing images unless otherwise mentioned.

(ii) rMC1 Cells. Cells were grown either in four-chamberedtissue culture glass slides (NUNC, Naperville, IL) andexposed to HG for 5 days as described [32]. Cells were fixedwith freshly prepared paraformaldehyde (4%) for 2 h in iceor 4◦C overnight, washed 10 min each with PB (3 times),and blocked with 5% horse serum in PB for 1 h at roomtemperature. Following a 30-minute wash with PBS, cellswere incubated with primary antibodies (1 : 100 dilutions)overnight at 4◦C in a humidified chamber. After washingwith PB, cells were further incubated with correspondingsecondary antibodies conjugated with Alexa Fluor 488 orAlexa 594 at 1 : 500 dilutions for 1 h at 37◦C in a darkenedhumidified chamber. The cell-associated fluorescence wasobserved in Olympus BX51 fluorescence microscope.

2.9. Cytochrome c Oxidase and ATP Assays. The mitochon-drial oxidative phosphorylation (OxPhos) process consists ofthe electron transport chain (ETC) and ATP synthase, andit provides more than 90% of cellular energy. The terminalenzyme of ETC is cytochrome c oxidase (CcO). CcO transferselectrons from cytochrome c to oxygen, which is reducedto water. In intact mammalian cells, this reaction is theproposed rate-limiting step of the ETC under physiologicalconditions [33]. Therefore, CcO is an ideal marker enzymeof OxPhos and was analysed in this study.

(i) CcO Activity. The oxygen consuming capacity of CcO-analyzed in a closed chamber equipped with a micro-Clark-type oxygen electrode (Oxygraph system, Hansatech,Norfolk, England) as previously described [34, 35]. Briefly,frozen cells were solubilized in 10 mM K-HEPES (pH 7.4),40 mM KCl, 1% Tween 20, 1 μM oligomycin, 1 mM PMSF,10 mM KF, 2 mM EGTA, and 1 mM Na vanadate. CcO activ-ity was measured in the presence of 20 mM ascorbate and byaddition of increasing amounts of cow heart cytochrome c.

Oxygen consumption was recorded on a computer and ana-lyzed with the Oxygraph software. Protein concentration wasdetermined with the DC protein assay kit (Bio-rad). CcO-specific activity is defined as consumed O2 (μM)/min/mgtotal protein.

(ii) ATP Assay. Determination of ATP levels using the bio-luminescent method was performed as described previously[34, 35]. ATP was released from rMC1 cells maintained in24 well plates using the boiling method. Two hundred fiftyμl TE buffer (50 mM Tris-Cl, pH 7.4, 4 mM EDTA) wasadded to the wells and scrapped off. Cells were immediatelytransferred to a boiling water bath for 4 min. Sampleswere put on ice and sonicated briefly and centrifuged. ATPconcentration was determined in the supernatant with anATP bioluminescence assay kit (Invitrogen) according to themanufacturer’s manual. Relative fluorescence units (RLUs)were detected in a Luminometer (Promega, Madison, WI).Experiments were performed in triplicates for each conditionand data were standardized to the protein concentrationusing the Biorad protein assay kit.

2.10. Gel-Shift Assay. The gel-shift or electrophoreticmobility-shift assay (EMSA) provides a simple and rapidmethod for detecting DNA-binding activity of proteins. Weinvestigated the sXBP1 and HIF-1α activity of rMC1 nuclearproteins by using a commercially available gel-shift assay kitfrom Signosis (Sunnyvale, CA) according to manufacturer’sinstructions and as described before [32]. After incubationat 16◦C for 30 min with biotin-labeled probes and nuclearextracts (5.0 μg protein) in a PCR machine, the protein-DNA complexes were subjected to 6.5% nondenaturingpolyacrylamide gels prepared with TBE gel formulation [32].The gels were run at 100 V for 1 h until the Bromophenolblue dye front is three quarters down the gel, using 0.5x TBErunning buffer. For competitive assays, excess cold probeswere added to the reaction. Streptavin-HRP and ECL wereused to detect and capture the reactive bands using a CellBioscience FluorChem E System (Santa Clara, CA).

2.11. Statistical Analysis. Results are expressed as means+/− SE. Student’s t-test or one-way ANOVA followed byBonferroni Post Hoc Test was used to compare differencesbetween treatment conditions [8, 9]. A preset P value of

Experimental Diabetes Research 5

0

4

8

12

16

TXNIP iNOS TLR4 P2X7R GSR

elat

ive

mR

NA

leve

l

4.55

9.19

2.21

4.043.25

1.55

(P < 0.05 versus normal, n = 6)

(P=ns)

Retinal gene expression indiabetes versus normal rat

IL−1β

(a)

Normal rat retina Diabetic rat retina (4 weeks)

GCL

INL

ONL

GFAP (green)

(b)

0

100

200

300

400

500

600

Rel

ativ

e m

RN

A le

vel

IL-1β

(n = 6)P = 0.04P = 0.04

Normal Diab +scr-siRNA

Diab +TXNIP TGS

Diabetic rat retina (4 weeks)

(c)

0

5

10

15

20

Rel

ativ

e m

RN

A le

vel

TNF-α

(n = 6 )(P = 0.3 )

Normal Diab +scr-siRNA

Diab +TXNIP TGS

Diabetic rat retina (4 weeks)

(d)

0

10

20

30

40

50

Rel

ativ

e m

RN

A le

vel

GFAP

Normal Diab +scr-siRNA

Diab +TXNIP TGS

Diabetic rat retina (4 weeks)

(n = 6 )P = 0.03P = 0.02

(e)

Figure 1: Diabetes induces TXNIP and pro-inflammatory gene expression and glia reactivity in the rat retina. (a) Messenger mRNA levelsfor TXNIP (4.55-fold ± 0.96) and pro-inflammatory IL-1β (9.19± 4.62), iNOS (2.22 ± 0.76), and pattern recognition receptors TLR4 (4.04± 1.43) and P2X7R (3.26 ± 0.73) are increased significantly (P < 0.05, n = 6) in the retina of diabetic rats (4 weeks) when compared withthe normal retina. GS is marginally increased (1.55 ± 0.29) but not significant (P = 0.3). (b) GFAP staining, a marker of gliosis, is alsoincreased radially throughout the neuroretina in the diabetic rat versus the normal retina, suggesting Muller glia activation. GCL: ganglioncell layer; INL: inner nuclear layer; and ONL: outer nuclear layer. (c–e) TXNIP knockdown by siRNA targeted to the promoter (RNAi TGS,[9]) reduces IL-1β and GFAP mRNA levels in the diabetic retina as compared to the scr-siRNA-treated diabetic rat retina. Under theseconditions (4 weeks of diabetes induction in rats), we did not see an increase in retinal TNF-α mRNA level.

On IHC, we observed that the increased pro-inflammatorygene expression is associated with enhanced GFAP staining,a marker for glia activation, in the entire length of theneural retina from diabetic rats when compared with thenondiabetic normal rat retina (Figure 1(b)). Specifically, inthe normal rat retina, GFAP staining is restricted to the innerlimiting membrane of Muller glia end-feet and ganglion

cell layer (GCL), however, in the diabetic rat retina, GFAPfilaments extend from GCL to inner nuclear (INL), outernuclear layer (ONL), and the inner and outer plexiformlayers suggesting retinal injury and inflammation. Knock-down of TXNIP by intravitreal injection of promoter-targeted siRNA (RNAi TGS) further reduces IL-1β andGFAP up-regulation significantly (P < 0.05, n = 6) in

6 Experimental Diabetes Research

the diabetic rat retina (Figures 1(c) and 1(e)). Under theseconditions, we did not observe an enhancement of TNF-α, another pro-inflammatory cytokine, in the diabetic retina(Figure 1(d)). TNF-α has previously been shown to induceat early at 1-2 weeks of STZ-induced diabetic rat retinasand then suppresses until late stages of diabetes durationwhere it may play a critical role in retinal inflammation andBRB breakdown [26]. Thus, the results demonstrate thatthere is an early inflammatory response and gliosis in thediabetic retina. Retinal Muller glia forms a close associationwith neuron and blood vessel and plays a critical functionin maintaining retinal homeostasis in health and disease.Therefore, we propose that Muller glia reacts to retinalinjury both in retinal vessels and neurons in diabetes andproduces pro-inflammatory and survival factors to restoretissue homeostasis.

3.2. HG Induces TXNIP Expression and Innate ImmuneResponses in rMC1 in Culture. We have shown previouslythat TXNIP expression and Muller glia activation occurearly in the diabetic retina [9]. So far, a role for TXNIP inMuller cell activation and pro-inflammatory gene inductionunder HG exposure has not been investigated. Therefore, weundertook a series of experiments to examine the temporalresponse of rMC1 to chronic hyperglycemia in culture. Weshow that HG exposure in rMC1 induces a significant (P <0.05, n = 6) increase in TXNIP mRNA expression (2–24 h)when compared with low glucose (5.5 mM, LG) indicated astime 0 control (Figure 2(a)). We have previously shown thatTXNIP mediates pro-IL-1β expression in primary SchwannCells via the NF-κB-dependent pathway [10]. Therefore, weexamined if TXNIP expression correlates with pro-IL-1βinduction in rMC1 under HG. As shown in Figure 2(b), HGincreases IL-1β mRNA expression at 2 and 4 h (P < 0.01)and then reduces at 24 h to the level of the control at 0time. Furthermore, nuclear NF-κB level is also enhanced byHG in rMC1 from 2 to 24 h (Figure 2(c)), which correlateswith activation of caspase-1 in rMC1 (Figure 2(d)). Activecaspase 1 is responsible for processing pro-IL-1β (34-kDa) toan active mature form of IL-1β (17-kDa).

We further examine whether the protein levels of TXNIPand pro-IL-1β protein are induced by HG and if the levelsof NLRP3 and caspase-1 are important for processing pro-IL-1β to its mature form in a time-dependent manner.TXNIP expression is low in rMC1 under LG while HGupregulates TXNIP (∼4-folds) significantly from 2 h to 5days (Figures 3(a) and 3(c)). Nuclear level of phosphorylatedp65 subunit of NF-κB increases at 2 and 4 h but reduces at24 h (Figure 3(b)); however, it elevates again at day 2 throughday 5 (Figure 3(d)). Pro-IL-1β is also increased at 2 and 4 hof HG exposure and reduces at 24 h, and again induces at3 days of HG (Figures 3(c) and 3(d)), which correlates withNLRP3 expression. Procaspase 1 level is lower at 2 and 4 hand upregulated at 24 h (Figure 3(c)). This is in agreementwith the increased level of active caspase-1 at 24 h shownin Figure 2(d). Pro-caspase-1 level rises again at 5 days ofHG exposure (Figure 3(d)). The results demonstrate thatTXNIP is a glucose-sensitive gene and its expression remainsupregulated when hyperglycemia persists. On the other

hand, the innate immune response to chronic hyperglycemiaby IL-1β, NLRP3 and caspase-1 inflammasome is cyclical.

3.3. Chronic Hyperglycemia Induces ROS Generation andReduces Cellular ATP in rMC1. We have observed previouslythat HG sustains TXNIP and induces IL-1β expression.TXNIP is a pro-oxidative stress protein through its inter-action/inhibition of the thiol reducing activity and ROSscavenging capacity of TRX. In addition, extracellular ATPrelease has been shown to involve in NLRP3 inflammasomeassembly, and IL-1β processing. Therefore, we examined atime-dependent generation of intracellular ROS and ATP aswell as extracellular ATP release in rMC1 upon HG exposure.HG significantly reduces ROS levels early at 4 and 24 h (P <0.05 versus LG, n = 6) and then rises at day 2 though notsignificant (Figure 4(a)). However, at day 3 of HG exposure,rMC1 cells produce significantly higher levels of ROS (P <0.05, n = 6) than LG and sustain up to day 5.

Excess nutrient (glucose availability) could enhanceglucose metabolic flux through glycolysis and mitochondrialOxPhos via the electron transport chain (ETC) and ATPproduction. Therefore, we measured temporal changes inintracellular ATP levels in rMC1 after HG exposure. Asshown in Figure 4(b), the intracellular ATP concentrationis significantly (P < 0.05, n = 4) enhanced at 4 h andreduces at 24 h when compared with time 0 control. Thereduction in ATP persists at day 2 through 5 (P < 0.01 versustime 0 control, n = 3). At day 3 and 5 of HG, the level ofATP is further reduced to ∼65% of the control ATP levelat LG (Figure 4(c)). In case of the extracellular ATP level,there is a small but significant increase at 4 and 6 h andthen at day 1 onwards their levels are reduced though notsignificant (unpublished data). These results indicate thatan initial increase in ATP generation via the mitochondrialOxPhos may result in ROS generation. Initially, this mayinduce an antioxidant response leading to a reduction in ROSlevel. However, as hyperglycemia persists, ROS overwhelmsthe anti-oxidant capacity.

3.4. Mitochondrial Electron Transport Chain (ETC) ProteinCytochrome c Oxidase Activity (CcO) and MTT Cell ViabilityAre Maintained under HG. As described previously, ROSgeneration and ATP reduction occur in rMC1 cells underchronic HG. Therefore, the mitochondrial ETC function,especially the oxygen consumption capacity of CcO activityat Complex IV, may be reduced. We measure the O2consuming capacity of CcO in an in vitro assay using aclosed oxygen chamber. In contrast to our prediction, theCcO activity is unaffected by chronic HG exposure of rMC1(Figures 5(a) and 5(b)). In addition, we did not observean initial sigmoidal rate of the CcO activity using differentprotein concentrations of cytochrome c in the O2 consumingassay, indicating that, under these experimental conditions,CcO may not be regulated by posttranslational modificationas previously described [34].

We next examined whether cell viability is reduced byROS and ATP in rMC1 cells using MTT assay. An increasein MTT absorbance measures the reductase activity ofmitochondrial succinate dehydrogenase at complex II, which

Experimental Diabetes Research 7

0

1

2

3

0 2 4

Rel

ativ

e m

RN

A le

vel

Treatment, HG

TXNIP

24 h

∗ ∗∗(

∗,P < 0.05 versus time 0 , n = 6 )rMC1, TXNIP

(a)

0

1

2

3

0 2 4

Rel

ativ

e m

RN

A le

vel

Treatment, HG

24 h

(∗,P < 0.01 versus time 0 , n = 6 )rMC1, IL-1β

IL-1β

∗

(b)

NF-κBp65

(red)

0 2 4 24 hHG

rMC1, nuclei-blue (DAPI)

(c)

0 2 4 24 hHG

rMC1, nuclei-blue (DAPI)

Active

caspase-1

(green)

(d)

Figure 2: HG induces TXNIP expression and evokes the host innate immune response in rat Muller glia. Quantitative RT-PCR for (a)TXNIP and (b) IL-1β mRNA expression in rMC1 is shown. HG increases TXNIP mRNA at 2–24 h (P < 0.05 versus time 0, n = 6) whilepro-IL-1β mRNA is increased at 2 and 4 h (P < 0.05 versus time 0, n = 6) and then reduces at 24 h. (c) IHC detects a time-dependentaccumulation of the p65 subunit of NF-κB in the nucleus in rMC1 under HG. (d) Active caspase-1 staining is increased in rMC1 using thecaspase-1 FLICA probe. A representative of n = 3 is shown here.

is frequently used to monitor cell viability. As shown inFigure 5(c), MTT activity is not affected by HG exposureup to 2 days and increases at day 3 through 5 (P < 0.01,n = 6). These results suggest that rMC1 maintains a func-tional mitochondria and viability in spite of the ROSgeneration and ATP depletion. To further ascertain that HGmaintains rMC1 viability under increased ROS, ATP release,and IL-1β expression, we treated rMC1 cells with exogenousH2O2 (0.2 mM), ATP (0.1 mM), and IL-1β (10 ng/mL) withor without HG. We observed that H2O2, ATP, and IL-1βincrease ROS levels significantly (P < 0.05, n = 6) at 4 to96 h and reduce cell viability as indicated by reduced MTTlevels under LG at 48 h (unpublished data). However, when

HG is added, the effect of H2O2, ATP, and IL-1β on MTTactivity is nullified. This occurs even when HG increasesthe generation ROS further with H2O2, ATP, or IL-1β incombination. Therefore, these findings suggest that rMC1activates a cellular defense and survival mechanism(s) underHG and oxidative stress, and TXNIP may play a role in thisprocess.

3.5. HG Increases Autophagy/Mitophagy, Inflammation, andApoptosis in rMC1 Cells. We observed that HG increasesTXNIP expression and ROS generation and sustained ATPreduction in rMC1 cells. Nonetheless, the mitochondrialCcO enzyme activity and cell viability are maintained.

8 Experimental Diabetes Research

0

2

4

6

8

10

0 2 0 24

Fold

ch

ange

0

2

4

6

8

Fold

ch

ange

HG HG

24 h

0 2 4 24 h

(P < 0.01,n = 4 )

50

43

50

43

67

34

45

106

50

67

34

45

106

50

(kDa) (kDa)

TXNIP

Actin

TXNIP

Actin

3 5 days

0 2 3 5 days

(P < 0.02,n = 4)

(a)

(b)

(c)

(d)

Pro-caspase-1

NLRP3

Tubulin

Pro-caspase-1

NLRP3

Tubulin

0 2 4 24 h 0 2 3 5 days

Pro-IL−1βPro-IL−1β

pNF-κBpNF-κB

Figure 3: Chronic hyperglycemia persistently upregulates TXNIP expression in rMC1 and induces IL-1β and NLRP3 inflammasomeactivation. rMC1 cells were cultured under HG for (a and b) 0–24 h or (c and d) 0–5 days and TXNIP proteins were detected by Westernblotting. For this, cell extracts were prepared in RIPA buffer and 30 μg protein was analyzed on 12% SDS-PAGE and Western blot forcytosolic TXNIP, Pro-IL-1β, NLRP3, and pro-caspase-1 and the nuclear level of phosphorylated p65 at serine residue 276 (S276) of NF-κB.ECL detected the immunoreactive bands. Actin and tubulin were used as controls for protein loading. A representative blot for each proteinis shown here from n = 3-4.

Therefore, we hypothesize that a cell survival program isevoked under chronic hyperglycemia and oxidative stressin rMC1. Excess ROS production causes protein misfold-ing/aggregation and organelle damage (e.g., mitochondria).Autophagy is a cellular survival mechanism for long-livedfully differentiated cells and for phagocytes under cellularstress to remove damage organelles and their regeneration[36, 37]. Beclin 1 is an autophagy initiating protein thatinteracts with prosurvival bcl2 and bcl-xL [38]. Underoxidative stress, beclin 1 is released from bcl2 and activatesthe autophagic process, which further activates the late phageautophagosome marker, light chain LC3B [39]. Therefore,we examined beclin 1 expression in rMC1 under chronic

hyperglycemia at day 3 and 5. As shown in Figure 6(a), beclin1 staining is marginally increased in rMC1 under HG than inLG, however there are more beclin 1 punctae in the cytosoland over the nuclei suggesting activation of an autophagicprocess. This beclin 1 puncta is not seen up to day 3 (notshown).

Next, we used a mitochondrial staining dye, MitoTrackerred, to examine the mitochondrial morphology. We observedan enhanced mitochondrial staining as well as the formationof larger mitotracker puncta (Figure 6(c)), indicating thatautophagy/mitophagy may occur in rMC1 under chronichyperglycemia. Subsequently, using an autophagy assay kitfor LC3B staining, we show that HG increases the number

Experimental Diabetes Research 9

0

20

40

60

80

100

120

140

160

LG 1 2 3 4

RFU

s (n

orm

aliz

ed to

LG

)

Treatment

(ns)

rMC1, ROS generation

(∗,P < 0.05 versus LG; n = 6)

∗

∗

HG, 4 h 5 days

(a)

0

20000

40000

60000

80000

100000

120000

RLU

s/u

g pr

otei

n

0 2 4 24 hHG

(∗P < 0.05 versus 0 time; n = 6 )rMC1, intracellular ATP

∗

∗

(b)

0

20000

40000

60000

80000

100000

120000

RLU

s/u

g pr

otei

n

rMC1, intracellular ATP

0 2HG

3 5 days

(P < 0.01;n = 3)

(c)

Figure 4: Time-dependent effect of HG on ROS and ATP generation in rMC1. (a) Reactive oxygen species (ROS) generation was measuredusing a fluorescence probe, CM-H2DFA. ROS level is reduced at both 4 and 24 h significantly (P < 0.05, n = 6) but rises at day 3 to 5(P < 0.05, n = 6). (b-c) ATP level was determined in rMC1 cells by a Renilla-based luminescence assay kit (Invitrogen). Intracellular ATP isincreased at 4 h (P < 0.05, n = 6) and reduces at 24 h as well as at day 2 through 5 (P < 0.01, n = 3). However, at day 3 and 5, the level ofATP is further reduced to ∼65% of the control.

and size of LC3B puncta formation (LC3BII, active form) inrMC1 when compared with LG (Figure 6(d)). Smaller LC3Bpunctae are also present in rMC1 after 5 days of LG underlow serum medium. In addition, we observed that at 8 weeksof diabetes, formation of LG3B puncta occurs in the retinaand they colocalize with TXNIP (unpublished data).

The mechanism of autophagy/mitophagy initiationunder chronic hyperglycemia, TXNIP, and ROS is not fullyunderstood at the present time. However, the thiol oxidationof the nuclear high mobility group box protein 1 (HMGB1)under oxidative stress has been implicated in its nuclearexport and cytosolic interaction with beclin 1 to release frombcl2 and to initiate an autophagic process [39]. Therefore, weexamined levels of TXNIP and HMGB1 in the nucleus andcytosol. We observed that TXNIP expression in the nucleusis also increased in rMC1 under HG (unpublished data) inaddition to their cytosolic expression as observed in Figure 3.

HMGB1 levels, both in the nucleus and in the cytosol, are notaltered at 0–24 h but decrease at day 2 and 3 in the cytosol andthen increase again at day 5. Furthermore, we observed thatprotein cysteine (thiol) nitrosylation (SNO) is enhanced inHG as measured by an anti-SNO antibody on IHC, and SNOcolocalizes with HMGB1 (unpublished data). In addition,beclin 1 and HMGB1 colocalize in the cytosol. These resultsshow that autophagy/mitophagy is activated in rMC1 underHG and that beclin 1 and oxidized HMGB1 may play a rolein initiating the process.

Autophagy/mitophagy is initially a mechanism for cellsurvival; however excessive and continuous autophagy mayremove vital proteins and organelles and induce cell apop-tosis [36, 37]. Therefore, we examined whether proapoptoticcaspase-3 is activated under chronic HG in rMC1. As shownin Figures 7(a) and 7(b), HG increases TXNIP and caspase-3 staining in rMC1 when compared with LG, suggesting

10 Experimental Diabetes Research

0

5

10

15

20

25

0 10 20 30

0 h2 h

4 h24 h

Cytochrome c (μM)

CcO

act

ivit

y (O

2μ

M/m

in/m

g pr

otei

n)

(a)

0

5

10

15

20

25

0

03

10 20 30

5 d

Cytochrome c (μM)

CcO

act

ivit

y (O

2μ

M/m

in/m

g pr

otei

n)

(b)

0

40

80

120

160

200

LG, 0 1 2 3 4

Treatment

5 daysHG, 4 h

(P =ns; n = 8)

(P < 0.01;n = 6)

20

60

100

140

180

MT

T A

570

nm

(%

of

LG)

(c)

Figure 5: HG effects on mitochondrial electron transport chain (ETC) enzyme activity in rMC1. (a-b) Cytochrome c oxidase (CcO) activitywas determined by increasing the amount of substrate cytochrome c in a polarographic method as described in Methods. CoO activity isdefined as consumed O2 [μM]/min/protein [mg]. Oxidative metabolism was increased after incubation in high glucose medium as was seenby an incremental increase of CcO activity at 24 h as well as at 3 and 5 day by ∼12%. Shown are representative experiments (n = 3; standarddeviation

Experimental Diabetes Research 11

rMC1, LG HG, 5 days

Beclin 1 (red); nuclei (blue, DAPI)

rMC1, LG HG, 5 days

MitoTracker Red

LC3B (red); nuclei (blue, DAPI)Nı̈ave IgG (red); nuclei (blue, DAPI)

(a)

(b)

(c)

(d)

Figure 6: HG activates autophagy/mitophagy in rMC1. Cells were grown in HG for 5 days under low serum conditions in a four-well slidechamber with media changing for every 48 h and 24 h for the final day. Cells were fixed and stained for (a) beclin 1 antibody and (b) anaı̈ve IgG for control antibody reaction. There are more beclin 1 puncta in HG than in LG. (c) MitoTracker Red staining also reveals moremitochondrial puncta in HG than in LG. (d) LC3B antibody staining shows an increased LC3B level and autophagic puncta formation inrMC1 under HG than in LG. The pictures are representative of 3 separate experiments.

continues to accumulate under sustained hyperglycemia,the anti-oxidant capacity is overwhelmed and an oxidativestress occurs at day 3 to 5 when XBP1 activity is reduced.Activation of other branches of UPR/ER stress (e.g., PERK-eIF-2 and ATF6) is yet to be investigated. Nonetheless,sustained ROS/RNS generation will cause excessive organelledamage and protein aggregation leading to apoptotic signalinduction (Figure 7(b)), which may involve ER stress medi-ated apoptotic signals, CHOP (transcription factor CEBPhomologous protein), and Bim (Bcl-2 interacting mediatorof cell death) expression [41]. Further studies are beingpursued to test this hypothesis.

Under persistent hyperglycemia, ROS, and sustained ATPdepletion (Figure 4(b)), a hypoxic-like response in rMC1may evoke and induce pro-inflammatory gene expressionfor iNOS, VEGF-A, and Cox-2 as seen in Figures 7(d)–7(f).Therefore, we measured the activity of hypoxia inducibletranscription factor (HIF) 1α in EMSA and the resultis shown in Figure 8(d). There is no increase in DNA-binding activity of HIF-1α in rMC1 up to day 3 by HG.However, at day 5 the HIF-1α DNA-binding activity isenhanced, which correlates positively with the expressionof its down stream gene target, vascular permeability, andproangiogenic VEGF-A (Figure 7(f)). These results suggeststrongly that chronic HG evokes an initially time-dependentearly innate immune and ER stress response in rMC1,followed by ATP reduction, ROS accumulation, and HIF-1α

activation leading, at least in part, to pro-inflammatory andproangiogenic gene expression.

3.7. Knockdown of TXNIP by siRNA Blocks ATP Reduction,ROS Generation, and LC3B Activation by HG in rMC1.Finally, we asked to what extent the sustained inductionof TXNIP in rMC1 under HG is responsible for metabolicdysregulation in rMC1. We used a transient transfectionmethod of siRNA to knock down TXNIP and assess its effecton ATP and ROS levels and LC3B expression in rMC1 underHG. First, we showed that HG-induced TXNIP proteinlevel is reduced by two different TXNIP siRNAs (siTXNIP1and siTXNIP3). As shown in Figure 9(a), siTXNIP3 gavea consistent suppression of TXNIP. Conversely, HG is stillable to induce TXNIP expression in scrRNA-transfected cells.Therefore, we used the scrRNA (control) and siTXNIP3in further studies. We observe that HG reduces ATP levelsignificantly (P = 5.2E − 05, n = 6) in scrRNA-transfectedcells, which is absent in siTXNIP3-transfected rMC1 cells(Figure 9(b)). In addition, HG-induced ROS generationin scrRNA-transfected cells is also blocked by siTXNIP3(Figure 9(c)). Furthermore, the staining of autophagicmarker LC3B is increased by HG in scrRNA-treated cells,which is reduced by siTXNIP3 (Figure 9(d)). The resultsdemonstrate that TXNIP mediates, at least in part, cellularoxidative stress, ATP reduction, and an autophagic responsein retinal Muller glia under chronic hyperglycemia.

12 Experimental Diabetes Research

TX

NIP

/DA

PI

(a)

Cas

pase

-3/D

AP

I

rMC1, LG HG (5 days)

(b)

0

10

20

30

40

50

60

0 1 3

Rel

ativ

e m

RN

A le

vel

HG

TXNIP

5 days

(P < 0.05 )

(c)

0

5

10

15

20

25

Rel

ativ

e m

RN

A le

vel

iNOS

0 1 3

HG

5 days

(P < 0.05 )

(d)

012345678

0 1 3

Rel

ativ

e m

RN

A le

vel

HG

Cox-2

(P = 0.05)

5 days

(e)

00.5

11.5

22.5

33.5

0 1 3 5 days

Rel

ativ

e m

RN

A le

vel

HG

VEGF-A(P = 0.03)

(f)

Figure 7: HG-induced TXNIP expression correlates with proapoptotic caspase-3 expression in rMC1. (a) IHC of TXNIP. Txnip staining isenhanced in rMC1 under HG exposure for 5 days, which correlates with increased (b) proapoptotic caspase 3 staining in similar durationof HG using a caspase-3 FLICA staining kit. A representative of n = 3 is shown here. (c–f) HG induces TXNIP and pro-inflammatory geneexpression in rMC1. Total RNA was isolated with TRIZOL and mRNA levels for (c) TXNIP, (d) iNOS, (e) Cox-2, and (f) VEGF-A weremeasured by qRT-PCR at various time periods of HG exposure (n = 3-4). P values were compared against respective controls at time 0.

Experimental Diabetes Research 13

00.5

11.5

22.5

33.5

0 4 24

Rel

ativ

e m

RN

A le

vel

HG

48 h

rMC1, sXBP1

(a)

0

1

2

3

4

0 1 3

Rel

ativ

e m

RN

A le

vel

HG

rMC1, sXBP1

5 d

(b)

sXBP1/probe complex

Free probe

Probe

Alone + biotinylated probe + cold probe

0 4 h 1 d 1 d3 d 5 d 0

rMC1, HG

(c)

Probe

Alone + biotinylated probe + cold probe

0 4 h 1 d 1 d3 d 5 d 0

rMC1, HG

HIF-1α/probe complex

(d)

Figure 8: HG induces an early ER stress and a later hypoxia-like response in rMC1. Messenger RNA for the spliced form of ER stress markerXBP1 (sXBP1) was measured in rMC1 at various time periods (a) 0–24 h and (b) 0–5 days after HG exposure by qRT-PCR There is nosignificant change in sXBP1 mRNA (p = ns, n = 3) when compared with respective controls at time 0. (c) The DNA-binding activity ofsXBP1 in rMC1 was further measured by EMSA using nuclear extracts (5 μg) and biotin-labeled DNA probes with or without a competitivecold DNA probe as described in Methods. Protein-DNA complexes were detected by streptavin-HPR and ECL. In the presence of competitivecold probe, the sXBP1 reactive band is abolished indicating the specificity of the sXBP1-DNA binding (last two lanes). Probe alone withoutthe nuclear extract was also run as a control (first lane). The sXBP1 activity is increased at 4 h and 1 day in rMC1 by HG and then returnsto the control level observed at day 0. (d) The DNA-binding activity of the hypoxic response factor HIF-1α in nuclear extracts was alsomeasured using a biotin-labeled HIF-1α binding consensus DNA probe. HIF-1α activity is not increased up to day 3 in rMC1 by HG butenhances at day 5. Competitive cold probes block the DNA-binding of HIF-1α, indicating the specificity of the binding assay. The images arerepresentative of n = 3 in both (c) and (d).

14 Experimental Diabetes Research

(kDa)

TXNIP

TXNIPLG HG LG HG LG HG

Actin

scrRNA siTXNIP1 siTXNIP3rMC1, siRNA-48 h, HG-4 h

50

43

(a)

0

20000

40000

60000

80000

100000

120000

140000

160000

HGLG LG HG

RLU

s/u

g pr

otei

n

ATP (ns)

Treatment, 3 days

scrRNA siTXNIP3

(P = 5.2E−05) (P = 3.2E−05,n = 6)

(b)

0

50

100

150

200

RFU

s

ROS

HGLG LG HG

Treatment, 3 days

scrRNA siTXNIP3

(P = 0.05 ) (ns, n = 6)

(c)

LG HG

scrRNA

siTXNIP3

LC3B (red); nuclei (blue, DAPI)

LC3B

(d)

Figure 9: TXNIP knockdown by siRNA blocks HG-mediated ATP reduction, ROS generation, and LC3B expression in rMC1. (a) Seventy toeighty percent confluent rMC1 cells were transfected transiently with an scramble siRNA (scrRNA, control) or with TXNIP mRNA-targetedsiRNAs-siTXNIP1 and siTXNIP3 (20 nM each) using HiPerfect transfection reagent in six-well plates or 60 mm culture plates in duplicates.After 6 h, complete media containing 5% serum were replaced and kept for 24 h. Media was subsequently changed to low serum media (0.2%serum) for 48 h. Afterwards, HG was added for 4 h and TXNIP amount was measured by Western blotting. We observe that siTXNIP3 givesa consistent suppression of TXNIP (∼70%) and used in further studies. A representative blot of n = 4 is shown here. Actin was used as acontrol for protein loading and siRNA specificity. (b) rMC1 cells were transiently transfected in 24-well plates with siTXNIP3 for 48 h andthen HG was added for 72 h in low serum media. Intracellular ATP concentration was measured and normalized to protein concentration(RLUs/μg protein). HG reduces ATP level in scrRNA-treated rMC1 cells while siTXNIP3 transfection blunted HG effects. (c) Similar toATP determination, we also measured ROS levels after 72 h of HG exposure in both scrRNA and siTXNIP3-transfected rMC1 with CM-H2DCFDA. HG increases ROS levels in scrRNA cells (P < 0.05, n = 6) versus LG but this effect was not observed in siTXNIP3 cells. (d)LC3B staining under HG exposure for 5 days is also reduced in siTXNIP3-transfected cells when compared with the scrRNA cells undersimilar duration of HG exposure (right panels). Under LG, LC3B staining is minimal in both scrRNA and siTXNIP3-treated rMC1 cells after5 days. A representative of n = 3 is shown here.

4. Discussion

Muller glia plays an important function in retinal tissuehomeostasis by taking up glucose and nutrients from theblood. It supplies metabolic products to neurons for properfunctioning and participates in uptake and detoxification ofneurotransmitters [5, 42]. Therefore, an injury to Mullerglia itself and sustained gliosis will produce proinflammatorycytokines and proangiogenic factors that may further affect

the retinal vasculature and neurons leading to ocular compli-cations of diabetes [43, 44]. TXNIP plays a critical role in reti-nal inflammation and apoptosis [9, 26, 45]. Here, we provideevidences that HG activates TXNIP in retinal Muller cells invitro as well as in vivo in the diabetic rat retina (this study,[9]). We demonstrate a temporal response of Muller glia tochronic HG exposure and programming of a series of innatehost defense mechanisms which involve (i) sustained TXNIPup-regulation, (ii) innate immune/ER stress responses—(iii)

Experimental Diabetes Research 15

ATP reduction, (iv) oxidative/nitrosative stress, (v) hypoxic-like response, (vi) autophagy/mitophagy, (vii) inflammationand, (viii) an apoptotic signal. Thus, a temporal and spatialsurvival response of Muller cells to chronic hyperglycemiaemerges and TXNIP may orchestrate some of these events.Definitely, these observations will have important implica-tions toward understanding the molecular basis for diseasedevelopment and progression of DR and in designing stage-specific therapeutic strategies to reduce retinal inflammationand neurovascular dysfunction in diabetes.

TXNIP is considered as a pro-oxidative stress, pro-in-flammatory, and proapoptotic protein under chronic hyper-glycemia, diabetes, and cellular stress [9, 11, 13, 26, 46].However, these deleterious effects of HG in cells are observedat later stages of sustained hyperglycemic exposure whileTXNIP is an early response gene to HG and tissue injury.Therefore, TXNIP may be considered as an early sensor ofmetabolic and cellular danger and a significant componentof the innate host defense mechanism(s) and wound healingprocesses [10, 15, 16]. In support of this hypothesis, werecently showed that TXNIP is responsible for S100B-RAGEaxis-induced expression of the innate immune responsemediator, pro-IL-1β, and secretion of mature IL-1β inprimary Schwann cells in culture and partial sciatic nerveinjury recovery [10]. Under these conditions, an increasein ROS generation was not observed [8, 10]. This findingis in agreement with our current data showing that HGinduces TXNIP and pro-IL-1β expression and activation ofNLRP3 inflammasome and caspase-1 at 4–24 h without ROSgeneration (Figures 2 and 3). In fact, this is also similarto the condition of chronic Granulomatous Disease whereimpaired ROS production due to a genetic defect does notaffect IL-1β secretion [47, 48].

In acute infection models using circulating monocytesand stromal macrophages, however the NLRP3 inflamma-some assembly, caspase-1 activation, and pro-IL-1β process-ing and secretion require ROS generation and ATP release[15, 49–51]. According to this model, during oxidativestress, ROS dissociates TXNIP bound to TRX. TXNIP thenbinds to NLRP3 to mediate NLRP3 inflammasome assemblycontaining ASC (Apoptotic Speck Protein Containing aCaspase Recruitment Domain (CARD)) and procaspase-1[15, 16]. This subsequently leads to caspase-1 autocleavageand activation. Activated caspase-1 then processes pro-IL-1β to its mature form and is then secreted extracellularly.IL-1β is a potent pro-inflammatory cytokine and, upon re-lease, induces several pro-inflammatory cytokines andchemokines. Therefore, IL-1β expression is highly regulatedat multiple steps including at least two signals—a primingand a processing/maturation—while the mechanism ofextracellular secretion is yet to be determined [52].

In our study, pro-IL-1β, NLRP3 inflammasome andprocaspase-1 levels oscillate at 4 h and day 3 of HG exposure.The first response at 4 h occurs in the absence of ROS releaseas discussed previously while the second response at day 3occurs under ROS/oxidative stress. We and others [10, 21]showed that TXNIP is responsible for pro-IL-1β priming/expression but the role of TXNIP in NLRP3 inflammasomeassembly and caspase-1 activation is yet to be resolved as

conflicting data exist at present [15, 21, 53]. Therefore,further studies will focus on the mechanism(s) of NLRP3inflammasome activation and pro-IL-1β processing and therole of TXNIP in this process in the diabetic retina and underchronic hyperglycemia.

Nutrient excess (glucose availability) leads to increaserates of glycolysis and energy (ATP) production, which is evi-dent from our data showing elevated intracellular ATP levelsat 4 h. Mitochondrial ATP synthesis produces ROS duringelectron transfer and OxPhos at ETC in the mitochondrialinner membrane. Furthermore, glucose and ATP induceincreased protein synthesis in the ER lumen and ER stressleading to a UPR response to enhance protein folding andanti-oxidant capacity via an induction of mitochondrial andcytosolic superoxide dismutases and catalase [40]. Indeed,we observed an activation of the IRE1/XBP1 pathway inrMC1 by HG, which is demonstrated by the enhanced DNA-binding activity of nuclear sXBP1 in EMSA (Figure 8(c)).sXBP1 is known to activate the transcription of anti-oxidantgenes including those stated above [40]. This may partlyexplain why we observed a decrease in ROS level at 4 and24 h of HG treatment in spite of TXNIP up-regulation andIL-1β expression. Nonetheless, as hyperglycemia persists, thesXBP1 activity reduces and the defensive program is switchedto a reduction in cellular ATP content, potentially in anattempt to minimize mitochondrial ROS generation throughthe aerobic respiration [54]. This may activate the cytosolicanaerobic glycolysis via an inhibition of pyruvate kinaseat TCA cycle and lactate production, which is a survivalresponse for phagocytic and cancer cells in a hypoxic envi-ronment during inflammation and tumor progression [55,56]. Even then, the ETC in mitochondrial inner membranecontinues to leak electrons and produces ROS at complexes Iand III [57]. This eventually will lead to accumulation of ROSsince TXNIP up-regulation can reduce the ROS scavengingcapacity of TRX. Such an event seems to occur at day 3 of HGexposure in rMC1s, where there is a second surge in IL-1βand NLRP3 expression is observed (Figure 3(d)). The surgemay be a second attempt to correct cellular metabolic defectsand induce cell survival. However, these responses causea further decline in cellular ATP level evoking a hypoxic-like response (HIF-1α activation) under hyperglycemia andnormoxia (Figure 8(d)).

Sustained TXNIP up-regulation may establish an oxida-tive stress environment in rMC1 at later stages of HG expo-sure (Figure 4(a)) after an initial innate immune response,ER stress, and ATP reduction. As ROS and iNOS expressionsincrease at day 3, nitrosative stress and protein thiol S-nitrosylation (SNO) as well as tyrosine nitration activitieswill be enhanced [58–60]. Initially, SNO modification of pro-apoptotic caspase-3 prevents its release from mitochondria[60]; however continued ROS/RNS will damage proteins,nucleic acids, and organelles including ER and mitochondria.Therefore, another ancient survival program of autophagy(mostly activated under nutrient starvation to recycle proteinaggregates and organelles) and mitophagy (removal ofdamaged mitochondria by autophagic processes) is activated[36–39]. This is supported by our observation that inspite of the significant increase in ROS levels (day 3 to 5

16 Experimental Diabetes Research

of HG), activity of mitochondrial membrane enzymecytochrome c oxidase at complex IV is sustained and cellviability is preserved. Preservation of cell viability is indicatedby increased MTT that measures the reductase activityof succinate dehydrogenase at complex II. These findingssuggest an active regeneration of mitochondria by removingthe oxidatively damaged ones and provide cell viability.Such an assumption is supported by the observation thatautophagic LC3B punctae are increased in rMC1 (Figure 6)and MTT activity is enhanced (Figure 5).

The mechanism of autophagy initiation is complex andrecent studies have pointed to a role of the nuclear DNA-binding protein HMGB1 under oxidative stress [60]. NuclearTXNIP may cause HMGB1 oxidation (S-nitrosylation) andnuclear export [61, 62]), which participates in beclin 1 disso-ciation from antiapoptotic bcl2 and initiation of autophagy[38, 39]. Oxidation of HMGB1 at cysteine residues 23, 45,and 106, probably by inhibition of TRX by TXNIP in thenucleus, causes its translocation to the cytosol and interac-tion with beclin 1 [38, 39]. HMGB1 is an abundant nuclearprotein considered to have a dual function: it acts (i) asa transcriptional activator in the nucleus by binding tolinker DNA between nucleosomes inducing DNA bendingand enhances transcription factor accessibility, and (ii) as apro-inflammatory cytokine (passively or actively secreted)by injured and dying cells as a danger signal and damage-associated molecular pattern (DAMP) molecule, which bindsto innate pattern recognition receptors (PRRs) such asTLR2, TLR4 and RAGE [20, 61]. (iii) In addition, HMGB1may also have a role in autophagy in rMC1 (this study,[60]). Previously, we showed that RAGE activation by itsendogenous ligand S100B activates TXNIP expression andinduces pro-inflammatory genes for IL-1β, Cox-2, VEGF-A, ICAM1, and fibrotic FN [8–10]. While IL-1β is an earlymediator of innate immunity and inflammation, HMGB1is regarded as a late mediator in chronic inflammatorydiseases [63, 64]. Thus, the HMGB1-TLR4 and/or HMGB1-RAGE axis may further activate TXNIP and IL-1β expressionthereby potentiating a vicious cycle of innate immunity andsterile inflammation under chronic hyperglycemia. The factthat knockdown of TXNIP by siRNA blocks IL-1β expressionin the retina in vivo (Figure 1(c)) and ATP depletion,ROS generation, and LC3B induction in rMC1 in vitro(Figure 9) suggest that TXNIP is critical for Muller gliaactivation, autophagy, and inflammation. TXNIP thereforemay represent a potent gene and therapeutic target to preventDR pathogenesis [8, 9].

Sustained ROS/RNS generation and activation of auto-phagy/mitophagy will eventually trigger an apoptotic signaland premature cell death due to excessive removal ofessential organelles and proteins in the cell while dam-aged mitochondria leaks electrons [57, 58]. Mitochondrialoxidative/nitrosative stress causes ER stress, which in turninduces mitochondrial damage thus generating a viciouscycle of organelle damage and tissue injury [41]. Here,we show that increases in both TXNIP and proapoptoticcaspase-3 occur at later periods of HG exposure for example,at day 5. Nonetheless, we propose that the ER stress andautophagic and apoptotic signal will be shifted towards a

caspase-1/7-dependent pro-inflammatory cell death (pyrop-tosis) [65]. This is because apoptosis is an ATP-dependent(consuming) process and requires active ATP synthesis [65].In the current study, we observe ATP depletion undersustained HG and a late surge in caspase-1 level at day5. Furthermore, the caspase-3 probe, FLICA, used in thisstudy may also recognize caspase-7, which is downstream ofcaspase-1 and closely related to caspase-3. In agreement withthe idea of pyroptosis under HG, the dying and oxidativelystressed rMC1 produces pro-inflammatory genes such asTXNIP, iNOS, Cox-2, and proangiogenic VEGF-A (Figures7(c)–7(f); [66]). The induction of VEGF-A is particularlyinteresting because of its association with ATP depletionand HIF-1α activation. VEGF-A is a potent vascular per-meability and proangiogenic factor that is regulated byHIF-1α and implicated in diabetic vasculopathy [56, 67].Protein S-nitrosylation also consumes molecular oxygen[58]; therefore, it may result in reduced mitochondrial ATPproduction and generation of an intracellular hypoxic-likeenvironment inducing HIF-1α activation under normoxia.An understanding of the TXNIP-HIF-1α-VEGF-A axis andits temporal response under chronic hyperglycemia in Mullercells that enhances VEGF-A expression in diabetes will beimportant. It is generally accepted that Muller glia is the mainsource of VEGF-A in the neural retina [68, 69].

An accumulation of HIF-1α in diabetic retinas of STZ-mice was recently reported [67], which is also associatedwith an increase in VEGF-A expression. Furthermore, adisruption of the Hif-1α gene in Müller cells alleviates retinalinflammation and vascular leakage in diabetic Hif-1α KOmice, further indicating an important role of Müller glia-derived HIF-1α-VEGF-A axis in DR [67]. In addition, atten-uation of ER stress by chemical chaperones such as 4-phenyl butaric acid reduces VEGF-A expression in retinalendothelial cells in culture induced by hypoxia as well as in amouse model of oxygen-induced retinopathy (OIR) [23, 70,71]. Thus, several reports have pointed to an important roleof ER stress/UPR signaling in retinal inflammation, hypoxia,and perhaps autophagy. Nonetheless, a critical function ofTXNIP in ER stress, autophagy, and apoptosis/pyroptosisin animal models of DR remains to be investigated, whichwill form the basis for a separate manuscript. Sustainedand excessive induction of each of these cellular defen-sive/survival processes, including ER stress and UPR, willevoke chronic inflammation and DR pathogenesis. ER-stress, autophagy, and innate immune responses are survivalmechanisms; however prolonged induction of these cellularprocesses will ultimately lead to premature cell death and dis-ease progression of microvascular complications of diabetes.Cellular DAMPs (ATP, uric acid, HMGB1, mitochondrialDNA, etc.) released by injured and dying cells are recognizedby innate PRRs such as TLR4, TLR9, RAGE, P2X7R, andNLRP3 inflammasome and together the DAMPs and PRRsmay play causative roles in retinal ER-stress, autophagy,inflammation, and DR pathogenesis.

In conclusion, we show for the first time that HG sustainsTXNIP expression in rat Muller glia and orchestrates aduration-dependent cellular program of innate host defenseand survival mechanisms that culminate in oxidative stress,

Experimental Diabetes Research 17

Muller Glia

Chronic hyperglycemia

TXNIP

Innate immune response

ER stress/UPR

Oxidative stress/nitrosative stress

ATP Reduction

Hypoxic-like response

autophagy/mitophagy

Inflammation/(ER stress) apoptosis/pyroptosis

Progression of

diabetic retinopathy

Figure 10: A schematic summary of potential cellular responsesby retinal Muller glia in chronic hyperglycemia and diabetes. Thesequence of molecular events that retinal Muller cells react tochronic hyperglycemia include (i) sustained upregulation of TXNIP,(ii) an initial innate immune and UPR response to excess glucosemetabolism and oxidative phosphorylation (ATP generation), (iii)oxidative stress (ROS/RNS generation) and a hypoxia-like responsethrough ATP reduction, (iv) an induction of an autophagic-mitophagic pathway, and (v) ER-stress and inflammation. Thesecellular responses constitute intrinsic cell survival/defense mecha-nisms, which, under chronic cell stress and injury, may promotepremature cell death and disease progression of DR.

ER stress, autophagy, inflammation, and cell death. Whiledetailed mechanisms are yet to be worked out, a temporalpattern emerges here as to how retinal Muller glia mightrespond to chronic hyperglycemia in diabetes. These include(i) sustained TXNIP up-regulation, (ii) an initial innateimmune and ER stress response to excess glucose metabolism(ATP generation), (iii) oxidative stress and a hypoxia-like response through ATP reduction, (iv) induction ofan autophagic-apoptosis pathway, and (v) inflammation. Atemporal response of Muller glia to chronic hyperglycemiaand potential molecular events are summarized in Figure 10.These findings clearly point to a crucial role of TXNIP in

Muller glia activation, oxidative/nitrosative, and ER stress,and sterile inflammation under chronic hyperglycemia andsuggest a potential gene and drug target for preventingneurovascular injury/cell death and pathogenesis of DR.Lastly, while our manuscript is in submission, an article cameout to demonstrate purinergic and glycemic induction ofTXNIP in rMC1 in culture [72], which also supports ourfindings in the present study.

Conflict of Interests

All authors declare that they have no conflict of interests forthis study.

Acknowledgments

Research grants from Mid-West Eye Bank, Michigan, andJuvenile Diabetes Research Foundation International to Dr.L. P. Singh are acknowledged. Research funding to theDepartment of Anatomy and Cell Biology for Core facilitiesby Grant no. P30 EY04068 from the National Eye Institute,the National Institutes of Health, is also acknowledged. Sup-ports from Research to Prevent Blindness to the Departmentof Ophthalmology are also acknowledged. Dr. K. D. Nantwiis supported by NIH Grant NS065397. Parts of this studywere presented at the 2011 Annual Meeting of ARVO at Ft.Lauderdale, FL, May 31–June 5.

References

[1] N. Cheung, P. Mitchell, and T. Y. Wong, “Diabetic retinopa-thy,” The Lancet, vol. 376, no. 9735, pp. 124–136, 2010.

[2] T. S. Kern and A. J. Barber, “Retinal ganglion cells in diabetes,”Journal of Physiology, vol. 586, no. 18, pp. 4401–4408, 2008.

[3] J. Tang and T. S. Kern, “Inflammation in diabetic retinopathy,”Progress in Retinal and Eye Research, vol. 30, no. 5, pp. 343–358, 2011.

[4] T. Frey and D. A. Antonetti, “Alterations to the blood-retinalbarrier in diabetes: cytokines and reactive oxygen species,”Antioxidants and Redox Signaling, vol. 15, no. 5, pp. 1271–1284, 2011.

[5] H. Xu, M. Chen, and J. V. Forrester, “Para-inflammation in theaging retina,” Progress in Retinal and Eye Research, vol. 28, no.5, pp. 348–368, 2009.

[6] J. A. Vincent and S. Mohr, “Inhibition of caspase-1/inter-leukin-1β signaling prevents degeneration of retinal capillariesin diabetes and galactosemia,” Diabetes, vol. 56, no. 1, pp. 224–230, 2007.

[7] F. Giacco and M. Brownlee, “Oxidative stress and diabeticcomplications,” Circulation Research, vol. 107, no. 9, pp. 1058–1070, 2010.

[8] L. Perrone, T. S. Devi, K. I. Hosoya, T. Terasaki, and L. P.Singh, “Thioredoxin interacting protein (TXNIP) induces in-flammation through chromatin modification in retinal cap-illary endothelial cells under diabetic conditions,” Journal ofCellular Physiology, vol. 221, no. 1, pp. 262–272, 2009.

[9] L. Perrone, T. S. Devi, K. I. Hosoya, T. Terasaki, and L. P.Singh, “Inhibition of TXNIP expression in vivo blocks earlypathologies of diabetic retinopathy,” Cell Death and Disease,vol. 1, no. 8, article e65, 2010.

18 Experimental Diabetes Research

[10] O. Sbai, T. S. Devi, M. A. B. Melone et al., “RAGE-TXNIP axisis required for S100B-promoted Schwann cell migration,fibronectin expression and cytokine secretion,” Journal of CellScience, vol. 123, no. 24, pp. 4332–4339, 2010.

[11] J. Chen, G. Saxena, I. N. Mungrue, A. J. Lusis, and A. Shalev,“Thioredoxin-interacting protein: a critical link betweenglucose toxicity and β-cell apoptosis,” Diabetes, vol. 57, no. 4,pp. 938–944, 2008.

[12] D. W. Cheng, Y. Jiang, A. Shalev, R. Kowluru, E. D. Crook,and L. P. Singh, “An analysis of high glucose and glucosamine-induced gene expression and oxidative stress in renal mesan-gial cells,” Archives of Physiology and Biochemistry, vol. 112, no.4-5, pp. 189–218, 2006.

[13] P. C. Schulze, J. Yoshioka, T. Takahashi, Z. He, G. L. King, andR. T. Lee, “Hyperglycemia promotes oxidative stress throughinhibition of thioredoxin function by thioredoxin-interactingprotein,” Journal of Biological Chemistry, vol. 279, no. 29, pp.30369–30374, 2004.

[14] M. T. Forrester, D. Seth, A. Hausladen et al., “Thioredoxin-interacting protein (Txnip) is a feedback regulator of S-nitrosylation,” Journal of Biological Chemistry, vol. 284, no. 52,pp. 36160–36166, 2009.

[15] R. Zhou, A. Tardivel, B. Thorens, I. Choi, and J. Tschopp,“Thioredoxin-interacting protein links oxidative stress toinflammasome activation,” Nature Immunology, vol. 11, no. 2,pp. 136–140, 2010.

[16] K. Schroder, R. Zhou, and J. Tschopp, “The NLRP3 inflam-masome: a sensor for metabolic danger?” Science, vol. 327, no.5963, pp. 296–300, 2010.

[17] B. Vandanmagsar, Y. H. Youm, A. Ravussin et al., “The NLRP3inflammasome instigates obesity-induced inflammation andinsulin resistance,” Nature Medicine, vol. 17, no. 2, pp. 179–188, 2011.

[18] S. L. Cassel and F. S. Sutterwala, “Sterile inflammatory re-sponses mediated by the NLRP3 inflammasome,” EuropeanJournal of Immunology, vol. 40, no. 3, pp. 607–611, 2010.

[19] D. Yao and M. Brownlee, “Hyperglycemia-induced reac-tive oxygen species increase expression of the receptor foradvanced glycation end products (RAGE) and RAGE ligands,”Diabetes, vol. 59, no. 1, pp. 249–255, 2010.

[20] G. P. Sims, D. C. Rowe, S. T. Rietdijk, R. Herbst, and A. J. Coyle,“HMGB1 and RAGE in inflammation and cancer,” AnnualReview of Immunology, vol. 28, pp. 367–388, 2010.

[21] T. B. Koenen, R. Stienstra, L. J. Van Tits et al., “Hyperglycemiaactivates caspase-1 and TXNIP-mediated IL-1β transcriptionin human adipose tissue,” Diabetes, vol. 60, no. 2, pp. 517–524,2011.

[22] Y. Zhou, J. Lee, C. M. Reno et al., “Regulation of glucose home-ostasis through a XBP-1-FoxO1 interaction,” Nature Medicine,vol. 17, no. 3, pp. 356–365, 2011.

[23] J. Li, J. J. Wang, and S. X. Zhang, “Preconditioning withendoplasmic reticulum stress mitigates retinal endothelialinflammation via activation of X-box binding protein 1,”Journal of Biological Chemistry, vol. 286, no. 6, pp. 4912–4921,2011.

[24] A. J. Barber, T. W. Gardner, and S. F. Abcouwer, “The signif-icance of vascular and neural apoptosis to the pathology ofdiabetic retinopathy,” Investigative Ophthalmology and VisualScience, vol. 52, no. 2, pp. 1156–1163, 2011.

[25] P. S. Chan, M. Kanwar, and R. A. Kowluru, “Resistance ofretinal inflammatory mediators to suppress after reinstitutionof good glycemic control: novel mechanism for metabolicmemory,” Journal of Diabetes and Its Complications, vol. 24,no. 1, pp. 55–63, 2010.

[26] G. V. Bixler, H. D. Vanguilder, R. M. Brucklacher, S. R.Kimball, S. K. Bronson, and W. M. Freeman, “Chronic insulintreatment of diabetes does not fully normalize alterations inthe retinal transcriptome,” BMC Medical Genomics, vol. 4,article 40, 2011.

[27] A. Bringmann, I. Iandiev, T. Pannicke et al., “Cellular signalingand factors involved in Müller cell gliosis: neuroprotective anddetrimental effects,” Progress in Retinal and Eye Research, vol.28, no. 6, pp. 423–451, 2009.

[28] V. P. Sarthy, S. J. Brodjian, K. Dutt, B. N. Kennedy, R. P.French, and J. W. Crabb, “Establishment and characterizationof a retinal Müller cell line,” Investigative Ophthalmology andVisual Science, vol. 39, no. 1, pp. 212–216, 1998.

[29] T. S. Devi, L. P. Singh, K.-I. Hosoya, and T. Terasaki, “GSK-3β/CREB axis mediates IGF-1-induced ECM/adhesion moleculeexpression, cell cycle progression and monolayer permeabilityin retinal capillary endothelial cells: implications for diabeticretinopathy,” Biochimica et Biophysica Acta, vol. 1812, no. 9,pp. 1080–1088, 2011.

[30] Y. Jiang, D. W. Cheng, E. Levi, and L. P. Singh, “IGF-1 in-creases laminin, cyclin D1, and P21Cip1 expression inglomerular mesangial cells: an investigation of the intracellularsignaling pathway and cell-cycle progression,” Journal ofCellular Biochemistry, vol. 98, no. 1, pp. 208–220, 2006.

[31] L. P. Singh, D. W. Cheng, R. Kowluru, E. Levi, and Y. Jiang,“Hexosamine induction of oxidative stress, hypertrophy andlaminin expression in renal mesangial cells: effect of the anti-oxidant α-lipoic acid,” Cell Biochemistry and Function, vol. 25,no. 5, pp. 537–550, 2007.

[32] L. P. Singh, Y. Jiang, and D. W. Cheng, “Proteomic identifi-cation of 14-3-3ζ as an adapter for IGF-1 and Akt/GSK-3βsignaling and survival of renal mesangial cells,” InternationalJournal of Biological Sciences, vol. 3, no. 1, pp. 27–39, 2007.

[33] C. Piccoli, S. Scacco, F. Bellomo et al., “cAMP controls oxygenmetabolism in mammalian cells,” FEBS Letters, vol. 580, no.18, pp. 4539–4543, 2006.

[34] I. Lee, A. R. Salomon, S. Ficarro et al., “cAMP-dependenttyrosine phosphorylation of subunit I inhibits cytochrome coxidase activity,” Journal of Biological Chemistry, vol. 280, no.7, pp. 6094–6100, 2005.

[35] M. Hüttemann, K. D. Nantwi, I. Lee, J. Liu, S. Mohiuddin,and T. Petrov, “Theophylline treatment improves mitochon-drial function after upper cervical spinal cord hemisection,”Experimental Neurology, vol. 223, no. 2, pp. 523–528, 2010.

[36] H. S. Jung and M. S. Lee, “Role of autophagy in diabetes andmitochondria,” Annals of the New York Academy of Sciences,vol. 1201, pp. 79–83, 2010.

[37] K. Nakahira, J. A. Haspel, V. A. K. Rathinam et al., “Autophagyproteins regulate innate immune responses by inhibitingthe release of mitochondrial DNA mediated by the NALP3inflammasome,” Nature Immunology, vol. 12, no. 3, pp. 222–230, 2011.

[38] R. Kang, K. M. Livesey, H. J. Zeh, M. T. Lotze, and D. Tang,“HMGB1: a novel Beclin 1-binding protein active in auto-phagy,” Autophagy, vol. 6, no. 8, pp. 1209–1211, 2010.

[39] D. Tang, R. Kang, K. M. Livesey et al., “Endogenous HMGB1regulates autophagy,” Journal of Cell Biology, vol. 190, no. 5,pp. 881–892, 2010.

[40] Y. Liu, M. Adachi, S. Zhao et al., “Preventing oxidative stress: anew role for XBP1,” Cell Death and Differentiation, vol. 16, no.6, pp. 847–857, 2009.

[41] J. D. Malhotra and R. J. Kaufman, “ER stress and its functionallink to mitochondria: role in cell survival and death,” Cold

Experimental Diabetes Research 19

Spring Harbor Perspectives in Biology, vol. 3, no. 9, Article IDa004424, 2011.

[42] D. Attwell, A. M. Buchan, S. Charpak, M. Lauritzen, B. A.Macvicar, and E. A. Newman, “Glial and neuronal control ofbrain blood flow,” Nature, vol. 468, no. 7321, pp. 232–243,2010.

[43] T. M. Curtis, R. Hamilton, P. H. Yong et al., “Müller glial dys-function during diabetic retinopathy in rats is linked to accu-mulation of advanced glycation end-products and advancedlipoxidation end-products,” Diabetologia, vol. 54, no. 3, pp.690–698, 2011.

[44] J. Wang, E. Xu, M. H. Elliott, M. Zhu, and Y.-Z. Le, “Müllercell-derived VEGF is essential for diabetes-induced retinalinflammation and vascular leakage,” Diabetes, vol. 59, no. 9,pp. 2297–2305, 2010.

[45] M. M. Al-Gayyar, M. A. Abdelsaid, S. Matragoon, B. A. Pillai,and A. B. El-Remessy, “Thioredoxin interacting protein is anovel mediator of retinal inflammation and neurotoxicity,”British Journal of Pharmacology, vol. 164, no. 1, pp. 170–180,2011.

[46] S. M. Tan, Y. Zhang, A. J. Cox, D. J. Kelly, and W. Qi, “Tranilastattenuates the up-regulation of thioredoxin-interacting pro-tein and oxidative stress in an experimental model of diabeticnephropathy,” Nephrology Dialysis Transplantation, vol. 26, no.1, pp. 100–110, 2011.

[47] F. L. Van De Veerdonk, S. P. Smeekens, L. A. Joosten etal., “Reactive oxygen species-independent activation of theIL-1β inflammasome in cells from patients with chronicgranulomatous disease,” Proceedings of the National Academyof Sciences of the United States of America, vol. 107, no. 7, pp.3030–3033, 2010.

[48] R. Van Bruggen, M. Y. Köker, M. Jansen et al., “Human NLRP3inflammasome activation is Nox1-4 independent,” Blood, vol.115, no. 26, pp. 5398–5400, 2010.

[49] B. K. Davis, H. Wen, and J. P. Ting, “The Inflammasome NLRsin immunity, inflammation, and associated diseases,” AnnualReview of Immunology, vol. 29, pp. 707–735, 2011.

[50] J. Segovia, A. Sabbah, V. Mgbemena et al., “TLR2/MyD88/NF-κB pathway, reactive oxygen species, potassium efflux activatesNLRP3/ASC inflammasome during respiratory syncytial virusinfection,” PLoS One, vol. 7, no. 1, Article ID e29695, 2012.

[51] F. Bauernfeind, E. Bartok, A. Rieger, L. Franchi, G. Núñez, andV. Hornung, “Cutting edge: reactive oxygen species inhibitorsblock priming, but not activation, of the NLRP3 inflamma-some,” Journal of Immunology, vol. 187, no. 2, pp. 613–617,2011.

[52] G. Y. Chen and G. Nuñez, “Sterile inflammation: sensing andreacting to damage,” Nature Reviews Immunology, vol. 10, no.12, pp. 826–837, 2010.

[53] S. L. Masters, A. Dunne, S. L. Subramanian et al., “Activationof the NLRP3 inflammasome by islet amyloid polypeptideprovides a mechanism for enhanced IL-1β in type 2 diabetes,”Nature Immunology, vol. 11, no. 10, pp. 897–904, 2010.

[54] V. Aas, N. P. Hessvik, M. Wettergreen et al., “Chronic hyper-glycemia reduces substrate oxidation and impairs metabolicswitching of human myotubes,” Biochimica et Biophysica Acta,vol. 1812, no. 1, pp. 94–105, 2011.

[55] H. Lall, K. Coughlan, and V. V. Sumbayev, “HIF-1α proteinis an essential factor for protection of myeloid cells againstLPS-induced depletion of ATP and apoptosis that supportsToll-like receptor 4-mediated production of IL-6,” MolecularImmunology, vol. 45, no. 11, pp. 3045–3049, 2008.

[56] V. Nizet and R. S. Johnson, “Interdependence of hypoxic andinnate immune responses,” Nature Reviews Immunology, vol.9, no. 9, pp. 609–617, 2009.

[57] J. Papaconstantinou and C. C. Hsieh, “Activation of senescenceand aging characteristics by mitochondrially generated ROS:how are they linked?” Cell Cycle, vol. 9, no. 19, pp. 3831–3833,2010.

[58] A. Garedew and S. Moncada, “Mitochondrial dysfunction andHIF1α stabilization in inflammation,” Journal of Cell Science,vol. 121, no. 20, pp. 3468–3475, 2008.

[59] L. Zheng and T. S. Kern, “Role of nitric oxide, superoxide,peroxynitrite and PARP in diabetic retinopathy,” Frontiers inBioscience, vol. 14, pp. 3974–3987, 2009.

[60] T. Nakamura and S. A. Lipton, “Redox modulation by S-nitrosylation contributes to protein misfolding, mitochondrialdynamics, and neuronal synaptic damage in neurodegenera-tive diseases,” Cell Death and Differentiation, vol. 18, no. 9, pp.1478–1486, 2011.

[61] D. Tang, R. Kang, H. J. Zeh III, and M. T. Lotze, “High-mobility group box 1, oxidative stress, and disease,” Antioxi-dants and Redox Signaling, vol. 14, no. 7, pp. 1315–1335, 2011.

[62] M. T. Forrester, D. Seth, A. Hausladen et al., “Thioredoxin-interacting protein (Txnip) is a feedback regulator of S-nitrosylation,” Journal of Biological Chemistry, vol. 284, no. 52,pp. 36160–36166, 2009.

[63] J. A. Nogueira-Machado, C. M. Volpe, C. A. Veloso, and M.M. Chaves, “HMGB1, TLR and RAGE: a functional tripod thatleads to diabetic inflammation,” Expert Opinion on Therapeu-tic Targets, vol. 15, no. 8, pp. 1023–1035, 2011.

[64] M. A. Van Zoelen, H. Yang, S. Florquin et al., “Role of toll-likereceptors 2 and 4, and the receptor for advanced glycation endproducts in high-mobility group box 1-induced inflammationin vivo,” Shock, vol. 31, no. 3, pp. 280–284, 2009.

[65] T. Bergsbaken, S. L. Fink, and B. T. Cookson, “Pyroptosis: hostcell death and inflammation,” Nature Reviews Microbiology,vol. 7, no. 2, pp. 99–109, 2009.

[66] Y. Du, V. P. Sarthy, and T. S. Kern, “Interaction between NOand COX pathways in retinal cells exposed to elevated glucoseand retina of diabetic rats,” American Journal of Physiology, vol.287, no. 4, pp. R735–R741, 2004.

[67] M. Lin, Y. Chen, J. Jin et al., “Ischaemia-induced retinal neo-vascularisation and diabetic retinopathy in mice with con-ditional knockout of hypoxia-inducible factor-1 in retinalMüller cells,” Diabetologia, vol. 54, no. 6, pp. 1554–1566, 2011.

[68] W. Eichler, H. Kuhrt, S. Hoffmann, P. Wiedemann, and A.Reichenbach, “VEGF release by retinal glia depends on bothoxygen and glucose supply,” NeuroReport, vol. 11, no. 16, pp.3533–3537, 2000.

[69] R. N. Frank, “Diabetic retinopathy,” New England Journal ofMedicine, vol. 350, no. 1, pp. 48–58, 2004.

[70] N. Badiola, C. Penas, A. Miñano-Molina et al., “Induction ofER stress in response to oxygen-glucose deprivation of corticalcultures involves the activation of the PERK and IRE-1pathways and of caspase-12,” Cell Death and Disease, vol. 2, no.4, article e149, 2011.

[71] U. Özcan, E. Yilmaz, L. Özcan et al., “Chemical chaperonesreduce ER stress and restore glucose homeostasis in a mousemodel of type 2 diabetes,” Science, vol. 313, no. 5790, pp. 1137–1140, 2006.

[72] K. E. Trueblood, S. Mohr, and G. R. Dubyak, “Purinergicregulation of high-glucose-induced caspase-1 activation inthe rat retinal Müller cell line rMC-1,” American Journal ofPhysiology, vol. 301, no. 5, pp. C1213–C1223, 2011.

Submit your manuscripts athttp://www.hindawi.com

Stem CellsInternational

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Disease Markers

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

Immunology ResearchHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Parkinson’s Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttp://www.hindawi.com